//
// Project: Disagreement in science: Missing women



clear all
version 15.1  



//
// set locals





//
// AER PREVIOUS PUBS FROM ALL AEA JOURNALS

// call data
use "${data}/output/aer_data_gender.dta", clear
drop if month=="May" & year!=2019  // exclude AEA papers and proceedings
drop if year==2020  
keep if (comment | research_article)
append using "${data}/output/all_aea_data_nogender.dta"  // include observations from all AEA journals to capture seniority
keep if journal=="American Economic Review" | type=="JOURNAL ARTICLE" | type=="PAPERS" | type=="REGULAR ARTICLE" | type=="REGULAR ARTICLES" | type=="Regular papers" | type=="SHORTER PAPERS"

// generate number of previous publications in the journal by author
bysort full_name (article_id_chronological): generate previous_pubs = _n
keep if journal=="American Economic Review"  // observations from the AER

// generate indicator of last author in article
egen last_author_id = max(author_id), by(article_id)
generate last_author = author_id==last_author_id

// analysis
local i=1
local coefficients`i' "2"
regress previous_pubs i.last_author i.year if last_author_id>1, vce(robust)
matrix m`i' = r(table)
scalar n`i' = e(N)
scalar rsq`i' = e(r2)
* p-vales
foreach k of numlist `coefficients`i'' {
	local p`i'_`k' = m`i'[4,`k']
}



//
// ASR

// call data
use "${data}/output/asr_data_gender.dta", clear
keep if (comment | research_article)

// generate number of previous publications in the journal by author
bysort full_name (article_id): generate previous_pubs = _n
*replace previous_pubs = 5 if previous_pubs>5 & previous_pubs<.

// generate indicator of last author in article
egen last_author_id = max(author_id), by(article_id)
generate last_author = author_id==last_author_id

// analysis
local i=2
local coefficients`i' "2"
regress previous_pubs i.last_author i.year if last_author_id>1, vce(robust)
matrix m`i' = r(table)
scalar n`i' = e(N)
scalar rsq`i' = e(r2)
* p-vales
foreach k of numlist `coefficients`i'' {
	local p`i'_`k' = m`i'[4,`k']
}



//
// JAMA

// call data
use "${data}/output/jama_pubmed_data_gender.dta", clear
drop if year==2020  
drop if year<2002  // full author names from PubMed not available
keep if comment | research_article
drop if article_with_etal
drop if strpos(full_name, "Fontanarosa")  // this JAMA editor appeared as first author of letters to the editor

// generate number of previous publications in the journal by author
bysort full_name (article_id): generate previous_pubs = _n
*replace previous_pubs = 5 if previous_pubs>5 & previous_pubs<.

// include only years with Comment & Response section, after counting number of publications in years before
drop if year<2013  // Comment & Response section started in July 2013
drop if year==2013 & month=="January"  // Comment & Response section started in July 2013
drop if year==2013 & month=="February"  // Comment & Response section started in July 2013
drop if year==2013 & month=="March"  // Comment & Response section started in July 2013
drop if year==2013 & month=="April"  // Comment & Response section started in July 2013
drop if year==2013 & month=="May"  // Comment & Response section started in July 2013
drop if year==2013 & month=="June"  // Comment & Response section started in July 2013 

// generate indicator of last author in article
egen last_author_id = max(author_id), by(article_id)
generate last_author = author_id==last_author_id

// analysis
local i=3
local coefficients`i' "2"
regress previous_pubs i.last_author i.year if last_author_id>1, vce(robust)
matrix m`i' = r(table)
scalar n`i' = e(N)
scalar rsq`i' = e(r2)
* p-vales
foreach k of numlist `coefficients`i'' {
	local p`i'_`k' = m`i'[4,`k']
}



//
// Nature

// call data
use "${data}/output/nature_data_gender.dta", clear
drop if year==2020
keep if comment | research_article

// generate number of previous publications in the journal by author
bysort full_name (article_id): generate previous_pubs = _n
*replace previous_pubs = 5 if previous_pubs>5 & previous_pubs<.

// generate indicator of last author in article
egen last_author_id = max(author_id), by(article_id)
generate last_author = author_id==last_author_id

// analysis
local i=4
local coefficients`i' "2"
regress previous_pubs i.last_author i.year if last_author_id>1, vce(robust)
matrix m`i' = r(table)
scalar n`i' = e(N)
scalar rsq`i' = e(r2)
* p-vales
foreach k of numlist `coefficients`i'' {
	local p`i'_`k' = m`i'[4,`k']
}



//
// PNAS 

// call data
use "${data}/output/pnas_data_gender.dta", clear
drop if full_name=="II" | full_name=="III" | full_name=="IV" | full_name=="Jr" | full_name=="Jr."  // erroneously scraped as separate author-article observations
drop if year==2020  // PNAS started comments in 2008
keep if comment | research_article

// generate number of previous publications in the journal by author
bysort full_name (article_id): generate previous_pubs = _n
*replace previous_pubs = 5 if previous_pubs>5 & previous_pubs<.

// generate indicator of last author in article
egen last_author_id = max(author_id), by(article_id)
generate last_author = author_id==last_author_id

// analysis
local i=5
local coefficients`i' "2"
regress previous_pubs i.last_author i.year if last_author_id>1, vce(robust)
matrix m`i' = r(table)
scalar n`i' = e(N)
scalar rsq`i' = e(r2)
* p-vales
foreach k of numlist `coefficients`i'' {
	local p`i'_`k' = m`i'[4,`k']
}




//
// Science

// call data
use "${data}/output/science_data_gender.dta", clear
drop if year==2020
keep if comment | research_article

// generate number of previous publications in the journal by author
bysort full_name (article_id): generate previous_pubs = _n
*replace previous_pubs = 5 if previous_pubs>5 & previous_pubs<.

// generate indicator of last author in article
egen last_author_id = max(author_id), by(article_id)
generate last_author = author_id==last_author_id

// analysis
local i=6
local coefficients`i' "2"
regress previous_pubs i.last_author i.year if last_author_id>1, vce(robust)
matrix m`i' = r(table)
scalar n`i' = e(N)
scalar rsq`i' = e(r2)
* p-vales
foreach k of numlist `coefficients`i'' {
	local p`i'_`k' = m`i'[4,`k']
}



//
// significance stars for regressions above
forvalues j=1/6 {
	foreach k of numlist `coefficients`j'' {
		if `p`j'_`k''<0.1 & `p`j'_`k''>=0.05 {
			scalar st`j'_`k'="*"
		}
		else if `p`j'_`k''<0.05 & `p`j'_`k''>=0.01 {
			scalar st`j'_`k'="**"
		}
		else if `p`j'_`k''<0.01 & `p`j'_`k''>=0.001 {
			scalar st`j'_`k'="***"
		}
		else if `p`j'_`k''<0.001 {
			scalar st`j'_`k'="****"
		}
		else {
			scalar st`j'_`k'=""
		}
	}
}

